KR970049419A - Method and apparatus for blind separation of delayed and filtered sources - Google Patents

Method and apparatus for blind separation of delayed and filtered sources Download PDF

Info

Publication number
KR970049419A
KR970049419A KR1019960064577A KR19960064577A KR970049419A KR 970049419 A KR970049419 A KR 970049419A KR 1019960064577 A KR1019960064577 A KR 1019960064577A KR 19960064577 A KR19960064577 A KR 19960064577A KR 970049419 A KR970049419 A KR 970049419A
Authority
KR
South Korea
Prior art keywords
channel
signal
signals
approximation
channels
Prior art date
Application number
KR1019960064577A
Other languages
Korean (ko)
Inventor
토크콜라 카시
Original Assignee
빈센트 비. 인그라시아
모토로라 인코포레이티드
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 빈센트 비. 인그라시아, 모토로라 인코포레이티드 filed Critical 빈센트 비. 인그라시아
Publication of KR970049419A publication Critical patent/KR970049419A/en

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/16—Sound input; Sound output
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H21/00—Adaptive networks
    • H03H21/0012—Digital adaptive filters
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2134—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on separation criteria, e.g. independent component analysis
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H21/00—Adaptive networks
    • H03H21/0012—Digital adaptive filters
    • H03H21/0025—Particular filtering methods
    • H03H2021/0034—Blind source separation

Abstract

소스 신호들의 성분들을 포함하고 있는 혼합 신호들로부터 소스 신호들을 복구시키는 방법, 채널 1과 2의 혼합 신호들은 채널 1과 2의 적응 가중치(adaptive weights)에 의해 곱해지게 되어 채널 1과 2의 곱 신호들(product signals)을 만든다. 채널 1의 필터된 피드백(feedback) 신호는 채널 2의 근사 신호(approximation signal)로부터 발생된다. 그리고 채널 2의 필터된 피드백 신호는 채널 1의 근사신호로부터 발생된다. 채널 1의 필터된 피드백 신호는 채널 1의 근사 신호를 발생시키기 위해 채널 1의 곱 신호에 더해지며, 채널 2의 필터된 피드백 신호는 채널2의 근사 신호를 발생시키기 위해 채널 2의 곱신호에 더해진다. 비선형 함수가 출력 신호들을 만들기 위해 상기 근사 신호들에 인가되어 진다. 적응 가중치들과 필터된 피드백 신호들은 출력 신호들의 엔트로피를 최대화시키기 위해 조정이 되어 진다.A method for recovering source signals from mixed signals containing components of the source signals, the mixed signals of channels 1 and 2 are multiplied by the adaptive weights of channels 1 and 2 such that the product signals of channels 1 and 2 Make product signals. The filtered feedback signal of channel 1 is generated from an approximation signal of channel 2. And the filtered feedback signal of channel 2 is generated from the approximation signal of channel 1. The filtered feedback signal of channel 1 is added to the product signal of channel 1 to generate an approximation signal of channel 1, and the filtered feedback signal of channel 2 is further added to the product signal of channel 2 to generate an approximation signal of channel 2. Become. A nonlinear function is applied to the approximation signals to produce output signals. Adaptive weights and filtered feedback signals are adjusted to maximize the entropy of the output signals.

Description

지연 및 필터된 소스들을 블라인드 분리하는 방법과 장치Method and apparatus for blind separation of delayed and filtered sources

본 내용은 요부공개 건이므로 전문내용을 수록하지 않았음As this is a public information case, the full text was not included.

제5도는 본 발명에 따르는 블라인드 소스 분리 시스템을 도시한 단순 블록도.5 is a simple block diagram illustrating a blind source separation system according to the present invention.

제6도는 본 발명에 따르는 블라인드 소스 분리 시스템을 도시한 단순 블록도.6 is a simple block diagram illustrating a blind source separation system according to the present invention.

Claims (4)

소스 신호들의 성분들을 포함하고 있는 동일한 다수의 혼합 신호들로부터 다수의 소스 신호들을 복구시키는 방법에 있어서, 채널 1의 곱신호를 만들기 위해 채널 1의 혼합신호에 채널 1의 적응 가중치를 곱하는 단계와, 채널 2의 곱 신호를 만들기 위해 채널 2의 혼합 신호에 채널 2의 가중치를 곱하는 단계와, 채널 2의 근사 신호에서 채널 1의 필터된 피드백 신호를 발생시키는 단계와, 채널 1의 근사신호에서 채널 2의 필터된 피드백 신호를 발생시키는 단계와, 채널 1의 근사 신호를 만들기 위해 채널1의 필터된 피드백 신호를 채널 1의 곱 신호에 더하는 단계와, 채널 2의 근사 신호를 만들기 위해 채널 2의 필터된 피드백 신호를 채널 2의 곱 신호에더하는 단계와, 채널 1과 2의 출력 신호들을 만들기 위해 비선형 함수를 채널 1과 2의 도 근사신호들에 인가시키는 단계와, 채널 1과 2의 출력신호들의 엔트로피를 최대화시키고, 제1소스 신호를 채널 1의 출력 신호로, 제2소스 신호를 채널2의 출력 신호로 복구시키기 위해, 채널 1의 적응 가중치, 채널 1의 필터된 피드백 신호, 채널 2의 적응 가중치와 채널 2의 필터된 피드백 신호를 조정하는 단계로 구성되어 있는 방법.CLAIMS 1. A method for recovering a plurality of source signals from a same plurality of mixed signals comprising components of source signals, the method comprising: multiplying a mixed signal of channel 1 by an adaptive weight of channel 1 to produce a multiplication signal of channel 1; Multiplying the mixed signal of channel 2 by the weight of channel 2 to produce a multiplication signal of channel 2, generating a filtered feedback signal of channel 1 in the approximation signal of channel 2, and channel 2 in the approximation signal of channel 1 Generating a filtered feedback signal of the channel, adding the filtered feedback signal of channel 1 to the product signal of channel 1 to produce an approximation signal of channel 1, and filtering the channel 2 to produce an approximation signal of channel 2 Adding a feedback signal to the product signal of channel 2, and applying a nonlinear function to the approximate signals of channels 1 and 2 to produce output signals of channels 1 and 2. And the adaptive weighting of channel 1, to maximize the entropy of the output signals of channels 1 and 2, and to recover the first source signal to the channel 1 output signal and the second source signal to the channel 2 output signal, Adjusting the filtered feedback signal of channel 1, the adaptive weight of channel 2 and the filtered feedback signal of channel 2. 소스 신호들의 성분들을 포함하고 있는 동일한 다수의 혼합 신호들로부터 다수의 소스 신호들을 복구하는방법에 있어서, 채널 1의 곱신호를 만들기 위해 채널 1의 혼합신호를 서로 다른 지연 시간과 관련된 채널 1의 서로 다른 가중치로 곱하고, 합하는 단계와, 채널 2의 곱 신호를 만들기 위해, 채널 2의 혼합신호를 서로 다른 지연 시간들과 관련된 채널 2의 서로 다른 가중치들로 곱하고, 합하는 단계와, 채널 2의 근사 신호를 지연시키고, 채널 2의 근사신호를 서로 다른 지연시간과 관련된 서로 다른 가중치로 곱하며, 합계하는 과정을 통해서, 채널 2의 근사 신호로부터 채널 1의 필터로 피드백 신호를 발생시키는 단계와, 채널 1의 근사 신호를 지연시키고, 채널 1의 근사 신호를 서로 다른 지연 시간을 갖고 있는 서로 다른 가중치들로 곱하며, 합계하는 과정을 통해서 채널1의 근사 신호로부터 채널 2의 필터된 피드백 신호를 발생시키는 단계와, 채널 1의 근사신호를 만들기 위해 채널1의 필터된 피드백 신호를 채널 1의 곱신호에 더하는 단계와, 채널 2의 근사 신호를 만들기 위해 채널 2의 필터된 피드백 신호를 채널 2의 곱신호에 더하는 단계와, 채널 1과 2의 출력 신호들을 만들기 위해, 비선형 합수를 채널 1과 2의 근사 신호들에 인가시키는 단계와, 채널 1과 2의 출력 신호들의 엔트로피를 최대화시켜, 채널 1의 소스 신호를 채널 1의 출력 신호로 그리고 채널 2의 소스 신호를 채널 2의 출력 신호로 복구시키기 위해서, 서로 다른 지연시간과 관련된 채널 1의 서로다른 가중치들과, 서로 다른 지연시간과 관련된 채널 2의 서로 다른 가중치들과, 채널 2의 필터된 피드백 신호를 조정하는 단계로 구성되어 있는 방법.A method for recovering a plurality of source signals from the same plurality of mixed signals containing components of the source signals, the method comprising: combining the mixed signals of channel 1 with each other of channel 1 with different delay times to produce a multiplication signal of channel 1; Multiplying by different weights and summating, multiplying and combining the mixed signals of channel 2 with different weights of channel 2 associated with different delay times to produce a multiplication signal of channel 2, and an approximation signal of channel 2 Generating a feedback signal from the approximation signal of channel 2 to the filter of channel 1 by multiplying and multiplying the approximation signal of channel 2 by different weights associated with different delay times, and channel 1 Delay the approximation of, multiply the approximation of channel 1 with different weights with different delay times, and sum Generating a filtered feedback signal of channel 2 from the approximation signal of channel 1 through adding the information, adding the filtered feedback signal of channel 1 to the product signal of channel 1 to produce an approximation signal of channel 1, and channel 2 Adding the filtered feedback signal of channel 2 to the product signal of channel 2 to produce an approximation of, and applying a nonlinear sum to the approximation signals of channels 1 and 2 to produce output signals of channels 1 and 2. And maximizing the entropy of the output signals of channels 1 and 2, so as to recover the source signal of channel 1 to the output signal of channel 1 and the source signal of channel 2 to the output signal of channel 2, Adjusting the different weights of channel 1, the different weights of channel 2 associated with different delay times, and the filtered feedback signal of channel 2 Law. 소스 신호들의 성분들을 포함하고 있는 동일한 다수의 혼합 신호들로부터 다수의 소수의 신호들을 복구시키는 방법에 있어서, 다음과 같이 채널 1의 근사 신호를 발생시키는 단계와,CLAIMS 1. A method for recovering a plurality of signals from the same plurality of mixed signals containing components of source signals, the method comprising: generating an approximate signal of channel 1 as follows: 다음과 같이 채널 2의 근사 신호를 발생시키는 단계와,Generating an approximate signal of channel 2 as follows; (W1k1, W2k2는 채널 1과 2에서 지연시간 k와 관련된 가중치들 W1k2, W2k1는 채널 1과 2에서 지연시간 k와 관련된 가중치들 u1,u2는 채널 1과 2의 근사신호들 x1,x2는 채널 1과 2의 혼합신호들) 채널 1과 2의 출력신호를 만들기 위해 비선형 함수를 채널1과 2의 근사 신호들에 인가시키는 단계와, 채널 1과 2의 출력 신호들의 엔트로피를 최대화시켜, 채널 1의 소스 신호를 채널 1의 출력신호로, 그리고 채널2의 소스 신호를 채널2의 출력 신호로 복구시키기 위해, W1k1, W2k2,W1k2, 와 W2k1을 조정하는 단계로 구성되어 있는 방법.(W 1 k 1 , W 2 k 2 are weights associated with delay k in channels 1 and 2 W 1 k 2 , W 2 k 1 are weights associated with delay k in channels 1 and 2 u 1 , u 2 Approximate signals of channels 1 and 2 x 1 , x 2 are mixed signals of channels 1 and 2) applying a nonlinear function to the approximate signals of channels 1 and 2 to produce an output signal of channels 1 and 2, and W 1 k 1 , W to maximize the entropy of the output signals of channels 1 and 2 so as to recover the source signal of channel 1 to the output signal of channel 1 and the source signal of channel 2 to the output signal of channel 2. A method consisting of adjusting 2 k 2 , W 1 k 2 , and W 2 k 1 . 제3항에 있어서, W1k1, W2k2,W1k2, 와 W2k1을 조정하는 단계는 다음 방정식에 따라 가중치들을 조정하는 단계를 포함하고 있는 방법.4. The method of claim 3, wherein adjusting W 1 k 1 , W 2 k 2 , W 1 k 2 , and W 2 k 1 comprises adjusting weights according to the following equation. △wioi∝(1-2yi)xi+1/wi0i,△ wioi ∝ (1-2 yi ) xi + 1 / wi0i , △wiki∝(1-2yi)xi(t-k),△ wiki ∝ (1-2 yi ) xi (tk), △wikj∝(1-2yi)uj(t-k), △ wikj α (1-2 yi) uj (tk), ※ 참고사항 : 최초출원 내용에 의하여 공개하는 것임.※ Note: The disclosure is based on the initial application.
KR1019960064577A 1995-12-12 1996-12-12 Method and apparatus for blind separation of delayed and filtered sources KR970049419A (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US08/571,329 US5675659A (en) 1995-12-12 1995-12-12 Methods and apparatus for blind separation of delayed and filtered sources
US571,329 1995-12-12

Publications (1)

Publication Number Publication Date
KR970049419A true KR970049419A (en) 1997-07-29

Family

ID=24283245

Family Applications (1)

Application Number Title Priority Date Filing Date
KR1019960064577A KR970049419A (en) 1995-12-12 1996-12-12 Method and apparatus for blind separation of delayed and filtered sources

Country Status (6)

Country Link
US (1) US5675659A (en)
EP (1) EP0779709A3 (en)
JP (1) JPH09252235A (en)
KR (1) KR970049419A (en)
CN (1) CN1097346C (en)
AU (1) AU705518B2 (en)

Families Citing this family (30)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
SE511496C2 (en) 1995-05-03 1999-10-11 Ulf Lindgren Mixed source signal separation method
US6185309B1 (en) * 1997-07-11 2001-02-06 The Regents Of The University Of California Method and apparatus for blind separation of mixed and convolved sources
US6167417A (en) * 1998-04-08 2000-12-26 Sarnoff Corporation Convolutive blind source separation using a multiple decorrelation method
US6526148B1 (en) * 1999-05-18 2003-02-25 Siemens Corporate Research, Inc. Device and method for demixing signal mixtures using fast blind source separation technique based on delay and attenuation compensation, and for selecting channels for the demixed signals
JP2001053654A (en) * 1999-08-16 2001-02-23 Matsushita Electric Ind Co Ltd Signal separating device, signal separation method and recording medium
EP1305975B1 (en) * 2000-06-13 2011-11-23 GN Resound A/S Adaptive microphone array system with preserving binaural cues
KR100394840B1 (en) * 2000-11-30 2003-08-19 한국과학기술원 Method for active noise cancellation using independent component analysis
US20040024596A1 (en) * 2002-07-31 2004-02-05 Carney Laurel H. Noise reduction system
KR20050115857A (en) * 2002-12-11 2005-12-08 소프트맥스 인코퍼레이티드 System and method for speech processing using independent component analysis under stability constraints
US20040117186A1 (en) * 2002-12-13 2004-06-17 Bhiksha Ramakrishnan Multi-channel transcription-based speaker separation
US7474756B2 (en) * 2002-12-18 2009-01-06 Siemens Corporate Research, Inc. System and method for non-square blind source separation under coherent noise by beamforming and time-frequency masking
US6954530B2 (en) * 2003-07-09 2005-10-11 Utah State University Echo cancellation filter
US7099821B2 (en) * 2003-09-12 2006-08-29 Softmax, Inc. Separation of target acoustic signals in a multi-transducer arrangement
KR100600313B1 (en) * 2004-02-26 2006-07-14 남승현 Method and apparatus for frequency domain blind separation of multipath multichannel mixed signal
US7231227B2 (en) * 2004-08-30 2007-06-12 Kyocera Corporation Systems and methods for blind source separation of wireless communication signals
WO2007028250A2 (en) * 2005-09-09 2007-03-15 Mcmaster University Method and device for binaural signal enhancement
WO2007103037A2 (en) * 2006-03-01 2007-09-13 Softmax, Inc. System and method for generating a separated signal
KR100875264B1 (en) 2006-08-29 2008-12-22 학교법인 동의학원 Post-processing method for blind signal separation
US20080155102A1 (en) * 2006-12-20 2008-06-26 Motorola, Inc. Method and system for managing a communication session
KR20090123921A (en) * 2007-02-26 2009-12-02 퀄컴 인코포레이티드 Systems, methods, and apparatus for signal separation
US8160273B2 (en) * 2007-02-26 2012-04-17 Erik Visser Systems, methods, and apparatus for signal separation using data driven techniques
EP2160716A1 (en) * 2007-06-08 2010-03-10 Thomson Licensing Method and apparatus for multi-lattice sparsity-based filtering
US8175291B2 (en) * 2007-12-19 2012-05-08 Qualcomm Incorporated Systems, methods, and apparatus for multi-microphone based speech enhancement
US8321214B2 (en) * 2008-06-02 2012-11-27 Qualcomm Incorporated Systems, methods, and apparatus for multichannel signal amplitude balancing
CZ303191B6 (en) * 2008-11-27 2012-05-23 Technická univerzita v Liberci Blind separation method of acoustic signals from convolution mixture thereof
US9099096B2 (en) * 2012-05-04 2015-08-04 Sony Computer Entertainment Inc. Source separation by independent component analysis with moving constraint
CN103236264B (en) * 2013-05-03 2015-07-29 湖北文理学院 A kind of decoding method for Sound seperation
CN109344754A (en) * 2018-09-21 2019-02-15 电子信息系统复杂电磁环境效应国家重点实验室 A kind of improvement type shortest path is deficient to determine source signal restoration methods
CN111696572B (en) * 2019-03-13 2023-07-18 富士通株式会社 Voice separation device, method and medium
CN117290708B (en) * 2023-11-24 2024-02-09 成都理工大学 High-low frequency aliasing signal separation method convenient for calculating transient impact quantity

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5208786A (en) * 1991-08-28 1993-05-04 Massachusetts Institute Of Technology Multi-channel signal separation
EP0565479A1 (en) * 1992-04-10 1993-10-13 Ramot University Authority For Applied Research & Industrial Development Ltd. Multi-channel signal separation using cross-polyspectra
US5383164A (en) * 1993-06-10 1995-01-17 The Salk Institute For Biological Studies Adaptive system for broadband multisignal discrimination in a channel with reverberation
US5500879A (en) * 1992-08-14 1996-03-19 Adtran Blind signal separation and equalization of full-duplex amplitude modulated signals on a signal transmission line
US5694474A (en) * 1995-09-18 1997-12-02 Interval Research Corporation Adaptive filter for signal processing and method therefor

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0623057Y2 (en) * 1990-03-15 1994-06-15 パイオニア株式会社 Crosstalk canceller circuit
US5237618A (en) * 1990-05-11 1993-08-17 General Electric Company Electronic compensation system for elimination or reduction of inter-channel interference in noise cancellation systems
JP3046203B2 (en) * 1994-05-18 2000-05-29 三菱電機株式会社 Hands-free communication device
US5515445A (en) * 1994-06-30 1996-05-07 At&T Corp. Long-time balancing of omni microphones
US5557682A (en) * 1994-07-12 1996-09-17 Digisonix Multi-filter-set active adaptive control system

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5208786A (en) * 1991-08-28 1993-05-04 Massachusetts Institute Of Technology Multi-channel signal separation
EP0565479A1 (en) * 1992-04-10 1993-10-13 Ramot University Authority For Applied Research & Industrial Development Ltd. Multi-channel signal separation using cross-polyspectra
US5500879A (en) * 1992-08-14 1996-03-19 Adtran Blind signal separation and equalization of full-duplex amplitude modulated signals on a signal transmission line
US5383164A (en) * 1993-06-10 1995-01-17 The Salk Institute For Biological Studies Adaptive system for broadband multisignal discrimination in a channel with reverberation
US5694474A (en) * 1995-09-18 1997-12-02 Interval Research Corporation Adaptive filter for signal processing and method therefor

Also Published As

Publication number Publication date
EP0779709A3 (en) 1998-01-21
JPH09252235A (en) 1997-09-22
AU705518B2 (en) 1999-05-27
CN1097346C (en) 2002-12-25
EP0779709A2 (en) 1997-06-18
AU7189696A (en) 1997-06-19
CN1154016A (en) 1997-07-09
US5675659A (en) 1997-10-07

Similar Documents

Publication Publication Date Title
KR970049419A (en) Method and apparatus for blind separation of delayed and filtered sources
KR920005507A (en) Digital signal encoder
Lee et al. Combining time-delayed decorrelation and ICA: Towards solving the cocktail party problem
CA2213348A1 (en) Method and apparatus for efficient implementation of single-sideband filter banks providing accurate measures of spectral magnitude and phase
MY114695A (en) Method and apparatus for reducing noise in speech signal
USRE40854E1 (en) Fast calculation apparatus for carrying out a forward and an inverse transform
US4071903A (en) Autocorrelation function factor generating method and circuitry therefor
KR860002812A (en) Nonlinear Signal Processing Device
US7822602B2 (en) Adaptive reduction of noise signals and background signals in a speech-processing system
US4574392A (en) Arrangement for the transmission of speech according to the channel vocoder principle
Singh et al. CAR-lite: A multi-rate cochlea model on FPGA
US4514853A (en) Multiplexed noise code generator utilizing transposed codes
US5627535A (en) Quantization apparatus
Fagerström et al. Dark velvet noise
KR970056177A (en) Lorentz Chaos Generation Circuit and Voice Communication System Using the Same
RU2024207C1 (en) Device to multiplex communicating channel
KR100863184B1 (en) Method for multichannel blind deconvolution to eliminate interference and reverberation signals
RU2022330C1 (en) Device for shaping systems of orthogonal digital signals
AU6502598A (en) Associative neuron in an artificial neutral network
SU571003A1 (en) Multichannel correlator
US5587910A (en) Vehicle navigation system with non-overflow digital filter
KR920004963A (en) Filtering Method in Transform Domain Using Pipeline Structure
SU1176455A1 (en) Method and apparatus for generating complex stereo signal
SU1043737A1 (en) Method and apparatus for forming signals controlling synthesis of speech
SU1195474A1 (en) Device for synchronous discrimination of phase-shift keyed signals

Legal Events

Date Code Title Description
A201 Request for examination
E701 Decision to grant or registration of patent right
NORF Unpaid initial registration fee